Learning PDE Time-Stepping with Neural Cellular Automata
Abstract
Classical numerical solvers for partial differential equations (PDEs) are computationally expensive to solve repeatedly across varying initial conditions, motivating the need for learned surrogates. In this paper, we propose a trainable Neural Cellular Automata (NCA) based surrogate model for learning long time PDE dynamics. Rather than mapping an entire initial field to a full trajectory in one shot, our proposed model learns a small, local, homogeneous update rule that is applied identically and repeatedly at every grid cell, mirroring the locality of differential operators. We benchmark this framework against three baselines: PDE - Net, a modified physics-informed neural network (PINN), and a Fourier Neural Operator (FNO), on five canonical PDEs (heat, advection, Burgers, Allen - Cahn, and Fisher - KPP), evaluated at temporal domain two times beyond the training temporal domain. The proposed model achieves the lowest long-horizon relative errors on the majority of the experiments.
Related
- Stable time-stepping via residual minimization: finite-element and neural-network approximations for transient parabolic problems
- Gardens of Eden and amenability on cellular automata
- Random Cellular Automata and deBruijn Graphs
- Predefined-Time Synchronization of Fractional-Order Delayed Memristor-Based Fuzzy Cellular Neural Networks
- When do lazy and invertible cellular automata generate all cellular automata?
BibTeX
@misc{saha2026learning,
title = {{Learning PDE Time-Stepping with Neural Cellular Automata}},
author = {Esha Saha and Hao Wang},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.30328},
eprint = {2608.30328},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.30328},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Learning PDE Time-Stepping with Neural Cellular Automata
AU - Esha Saha
AU - Hao Wang
PY - 2026
DA - 2026/08/31
PB - arXiv
DO - 10.48550/arxiv.2608.30328
UR - https://arxiv.org/abs/2608.30328
AB - Classical numerical solvers for partial differential equations (PDEs) are computationally expensive to solve repeatedly across varying initial conditions, motivating the need for learned surrogates. In this paper, we propose a trainable Neural Cellular Automata (NCA) based surrogate model for learning long time PDE dynamics. Rather than mapping an entire initial field to a full trajectory in one shot, our proposed model learns a small, local, homogeneous update rule that is applied identically and repeatedly at every grid cell, mirroring the locality of differential operators. We benchmark this framework against three baselines: PDE - Net, a modified physics-informed neural network (PINN), and a Fourier Neural Operator (FNO), on five canonical PDEs (heat, advection, Burgers, Allen - Cahn, and Fisher - KPP), evaluated at temporal domain two times beyond the training temporal domain. The proposed model achieves the lowest long-horizon relative errors on the majority of the experiments.
ER -